BCBS 239 is a standard set by the Bank for International Settlements which applies to banks which are systemtically important, either on a global or national basis. It raises the bar for IT systems and the challenge of compliance is a major concern of both IT organizations and the C-Suite.
A product of years of development at Ontology2, Real Semantics could be the first IT architecture designed with BCBS 239 in mind. With a universal graph model, a common mechanism can be used for tracing any decisions made by the system. With the capability to peer into legacy systems at the conceptual, API, object code and source code level, Real Semantics synchronizes the map and the territory to take the chaos out of IT, setting the stage for intelligent applications that satisfy customers and grow the top line.
In this chapter, we will look at a few paragraphs from the official BCBS 239 document and explain what features and characteristics of Real Semantics satisfy their requirements.
33. A bank should establish integrated 16 data taxonomies and architecture across the banking group, which includes information on the characteristics of the data (metadata), as well as use of single identifiers and/or unified naming conventions for data including legal entities, counterparties, customers and accounts.which comes with the associated footnote:
16 Banks do not necessarily need to have one data model; rather, there should be robust automated reconciliation procedures where multiple models are in use.
Real Semantics is based on a Multiple Model Architecture, which is precisely intended to interconnect systems based upon different data models. This is possible because the RDF/K data model is flexible to itself model any model in use by legacy systems. A single data model also supports a single model for metadata about every concept, data record, user, IT artifact or other thing encountered.
Ontology2 has supported an ongoing program to improve the quality and state of knowledge about Legal Entity Identifiers. As such, Real Semantics is ideal for reconciliation of various numeric identifiers as well as natural languages names and addresses.
34. Roles and responsibilities should be established as they relate to the ownership and quality of risk data and information for both the business and IT functions. The owners (business and IT functions), in partnership with risk managers, should ensure there are adequate controls throughout the lifecycle of the data and for all aspects of the technology infrastructure. The role of the business owner includes ensuring data is correctly entered by the relevant front office unit, kept current and aligned with the data definitions, and also ensuring that risk data aggregation capabilities and risk reporting practices are consistent with firms’ policies.
Real Semantics traceability makes control certain.
Real Semantics captures knowledge from workers across the organization from line workers to subject matter experts, software developers, risk management experts and executive. With a structured process for case management, misunderstandings can be understood, identified, repaired and/or documented. Policy documents are captured directly into the system, linked to rule-based implementations accessible to auditors and other professionals.
37. As a precondition, a bank should have a "dictionary" of the concepts used, such that data is defined consistently across an organisation.
The RDF/K model used in Real Semantics not only represents common data structures used in the software industry, but it works together with RDFS, OWL, SKOS and similar vocabularies to represent concepts used in natural language and cognition. Real Semantics exceeds traditional tools in that it keeps track of multiple contexts (Multiple Model Architecture), setting the stage to record nuances of expression that are are relevant to various communities, applications and business lines.
38. There should be an appropriate balance between automated and manual systems. Where professional judgements are required, human intervention may be appropriate. For many other processes, a higher degree of automation is desirable to reduce the risk of errors.
Real Semantics pushes the envelope of what can be automated because it is based on a systematic approach to capturing natural language documentation, implicit knowledge in the form of rules, and explicit knowledge in the form of examples which can be used to validate rules, explain special cases, and train machine learning models.
Human input is important, so Real Semantics participates in business process orchestration, sending difficult tasks to humans as well as sending a statistical sample for review to support statistical quality control (think Edward Deming, Phil Crosby, and Joseph Juran) to control, understand and improve quality with a focused and efficient use of human effort.
39. Supervisors expect banks to document and explain all of their risk data aggregation processes whether automated or manual (judgement based or otherwise). Documentation should include an explanation of the appropriateness of any manual workarounds, a description of their criticality to the accuracy of risk data aggregation and proposed actions to reduce the impact.
Experience shows that the structure of commonsense knowledge consists of rules and exceptions. In conventional software development based on procedural or object-oriented techniques, the result is often that a clean design becomes corrupted over time with a large number of "hacks" put in at the last minute that are placed randomly throughout the code base. The result is frequently a large system that is unpredictable, unreliable, and hard to maintain.
Real Semantics is based on the assumption that simple models are imperfect, so it provides facilities, such as fact patches and rule patches to create neat layers that separate general purpose, industry specific, company specific and department specific differences. Changes are documented with a case management system, so that clear documentation exists for both the automatic fast path as well as cases that are specialized or require human intervention.
40. Supervisors expect banks to measure and monitor the accuracy of data and to develop appropriate escalation channels and action plans to be in place to rectify poor data quality
Real Semantics flags quality problems automatically based upon both rules and discovery algorithms. In cases where data accuracy is in question, a ticket is created in the integrated case management system, so the problem can be investigated, possibly confirmed, and referred to the proper person, with the proper knowledge and skills, inside or outside of the organization -- leaving a paper trail which can be referred to later.
49. Adaptability includes: (a) Data aggregation processes that are flexible and enable risk data to be aggregated for assessment and quick decision-making; (b) Capabilities for data customisation to users’ needs (eg dashboards, key takeaways, anomalies), to drill down as needed, and to produce quick summary reports; (c) Capabilities to incorporate new developments on the organisation of the business and/or external factors that influence the bank’s risk profile; and (d) Capabilities to incorporate changes in the regulatory framework.
Rapid and flexible processing is enabled in Real Semantics by an extensive knowledge base about real world data and the problem domain. By bundling these capabilities into a structured process, the data cleaning that takes 80 to 90% of the time of a typical analyst or "data scientist" is built into a repeatable process that dramtically saves time.
Although Real Semantics can export data for use with any tools, it integrates with Kibi from our partner SIREn Solutions, which combines classical BI queries with time series and full-text capabilities. Real Semantics classifies documents, data records, and events, accurately extracting facts and concepts for an unmatched capability for data exploration and dashboard creation.
With the Multiple Model Architecture, Real Semantics is designed from square one with the assumption that data, code, documentation and other assets will be repurposed in multiple ways. With data lake capabilities based on Hadoop and cloud computing, Real Semantics can get the resources to repeat and reproduce data processing operations in case of any change of the data, rules, or other assumptions about the environment, business organization or regulatory frameworks quickly.
50. Supervisors expect banks to be able to generate subsets of data based on requested scenarios or resulting from economic events. For example, a bank should be able to aggregate risk data quickly on country credit exposures 18 as of a specified date based on a list of countries, as well as industry credit exposures as of a specified date based on a list of industry types across all business lines and geographic areas.
Real Semantics incorporates reasoning over large Linked Data knowledge bases covering topics such as places, people, creative works, industry classifications as well as technology, chemistry, biology and other fields that concern large numbers of creative entities. Real Semantics comes with the Ontology2 Spatial Hierarchy, derived from Freebase (which Google bought to create the Google Knowledge Graph).
Containing spatial relationships between millions of locations, as well in names in nearly 100 languages, the Ontology2 Spatial Hierarchy is a global database that places business locations in postal codes, cities, countries and other spatial subdivisions (such as US states and counties and Japanese prefectures) to identify regions and regulatory jurisdictions to support subreporting as necessary.
Many financial organizations see BCSB 239, Dodd-Frank, EMIR II and other regulations as a source of cost and risk. We see it is a wake-up call to fix creaking IT infrastructures: while many organizations are struggling to meet regulatory requirements, we see this as an opportunity to find alpha, satisfy and delight customers, and grow the bottom line. If you agree, contact us.